Douglas; it may very well be marketing babble more than actual technology.
I do find that the GXR files are very nice right out of the camera and through RPP. They are so crisp I sometimes hesitate to sharpen them in post. And they pop quite nicely with very minor work. Color is usually dead on, and all in all it is the best digital camera to date that I have worked with when it comes to post processing. Most of the time I am awed by the color accuracy and pop. I write down most of what I see to the well balanced sensor and lack of AA-filter. For high ISO there are certainly better options and for this reason I would welcome an A16 M module with the Sony sensor, sans AA filter. In the mean time I am very pleased with the A12 M module as it is.
You can add a lil' grain to the GXR files too, they look very film-ish if you do. Mostly out of RPP I find the files are quite true to film looks without further enhancement, without the drawbacks of film.
Some people don't like RPP at all but considering their accurate profile for the GXR and the easy workflow once set up I find it faster than other RAW converters and I get to where I want with a minimum of fuss, with the look that I want.
On a side note I am far less impressed with Fuji colors so far. Sure, I only use the X10 but colors are among the least accurate and pleasing of the digital cameras I have used so far. I sure hope the X Pro 1 will do a lot better. I found the Sony NEX5n and the Sony A900 easy to work with, pretty much as good as the GXR M mount and when I had the Olympus E3 I found it very accurate too.
I'm not so skeptical regarding the micro lenses question. There's something in it.
If the AA filter was the only factor for poor corners, how do you explain that the Sony N5N beats both the old N5 and the new N7 not only on color vigneting but on corner sharpness as well ?. All three NEX models have AA and IR filters, in all probability very alike regarding thickness.
My opinion is that micro lenses arrangement and pixel density play also a role.
artur5 wrote:
I'm not so skeptical regarding the micro lenses question. There's something in it.
If the AA filter was the only factor for poor corners, how do you explain that the Sony N5N beats both the old N5 and the new N7 not only on color vigneting but on corner sharpness as well ?. All three NEX models have AA and IR filters, in all probability very alike regarding thickness.
My opinion is that micro lenses arrangement and pixel density play also a role.
IR and AA filter types, distance of those filters to the sensor, CFA types, and pixel size all play a factor in astigmatism and color shift. You'll notice that Zeiss doesn't even mention special microlenses when talking about ZM lenses on digital in their wide angle paper, but, rather, sensor filters. Even with its "special" microlenses, just about any current CMOS has a better fill factor at the edges compared to the M9. I guess we need someone to rip off a Nex AA filter to see. Joakim probably knows the deal with the GXR microlenses. We should ask him.
kosmoskatten wrote:
Douglas; it may very well be marketing babble more than actual technology.
I do find that the GXR files are very nice right out of the camera and through RPP. They are so crisp I sometimes hesitate to sharpen them in post. And they pop quite nicely with very minor work. Color is usually dead on, and all in all it is the best digital camera to date that I have worked with when it comes to post processing. Most of the time I am awed by the color accuracy and pop. I write down most of what I see to the well balanced sensor and lack of AA-filter. For high ISO there are certainly better options and for this reason I would welcome an A16 M module with the Sony sensor, sans AA filter. In the mean time I am very pleased with the A12 M module as it is.
You can add a lil' grain to the GXR files too, they look very film-ish if you do. Mostly out of RPP I find the files are quite true to film looks without further enhancement, without the drawbacks of film.
Some people don't like RPP at all but considering their accurate profile for the GXR and the easy workflow once set up I find it faster than other RAW converters and I get to where I want with a minimum of fuss, with the look that I want.
On a side note I am far less impressed with Fuji colors so far. Sure, I only use the X10 but colors are among the least accurate and pleasing of the digital cameras I have used so far. I sure hope the X Pro 1 will do a lot better. I found the Sony NEX5n and the Sony A900 easy to work with, pretty much as good as the GXR M mount and when I had the Olympus E3 I found it very accurate too.
Don't get me wrong, I'd trade my 5N sensor for a GXR sensor for better edge performance. I'm just saying that, in the rest of the image, removing the AA filter creates digital artifacts that give the impression of more resolution, which isn't technically the case. Optimizing deconvolution sharpening and maybe adding a little grain will "improve" AA filtered cameras in the same way. Granted, many seem to prefer the look of AA-less cameras, because of this false detail, but it is achievable with other cameras, and there's a reason the manufacturers still use AA filters.
Yes, I can agree to most of that. (I'll have to let it sink in before I can agree to all of it ).
Maybe the tweaking done with the GXR M that I am so fond of has its attributes in the "sum of all parts" instead of just the sensor. I find the Sony 16MP sensor to be a very good balance of high ISO, dynamic range and resolution. That is why I am keen on seeing an AA-filterless A16 M mount hit the market. If it was just down to sheer MP count I would be quite happy with 12MP on APS-C.
I honestly wouldn't give a crap if the sensor has an AA-filter or not, as long as it is capable and the AA-filter doesn't exact a heavy toll on the "snap, crackle and pop" we aim for.
I am just curious as to what Ricoh has done to the sensor to make it carry a tune the way it does.
Again, don't get me wrong. I'd swap sensors with ya I just want to let you know that the snap, crackle, pop is partially due to digital artifacts creating false detail. When Sam sent me raws of the same scene with the same lenses on the GXR and 5N, a little tweak in LR on the 5N file yielded pretty similar results, accept at the edges, which had a little more resolution and less vignetting with the GXR file.
Despite wanting to rip off my AA on my Nex for edge reasons, the myth of more clarity on a AA-less camera needs to be dispelled a bit. I've owned a digital back before, so I know it can be initially mesmerizing.
Joakim "theSuede" is the real guy to talk to about this stuff.
Let's wait and see what happens with the Fuji X-pro 1 and M wides.
It will be enlightening to know how it compares with the NEX-5N in the corners. Almost same sensor but AA filter vs No AA filter.
Douglas, you should persuade Joakim to brief a bit Sean Reid about those matters. As much I appreciate some of Sean's reviews, I'm afraid that concerning optic science, he's as clueless as me..
Hmm, I can only go on what I have seen myself when using my RAW converters and in my work flow. I don't see those artifacts producing false detail that you are referring to. Maybe they are a byproduct of the RAW converter used? On some product shots I have made I was happy to see a real gain in micro detail, and not an artificial one. The gain is small, mind you, but it is there and the difference is tangible in the sense that the GXR files require less sharpening in post and clarity definitely is better where as the NEX5n required a bit more of both sharpening and local contrast enhancement - and that was introducing artifacts on a larger scale.
I think the benefit holds true when comparing same format or sensors of the same type and "magnitude"; full frame vs full frame, aps-c vs aps-c. YMMV of course and we can happily agree to disagree on that.
Perhaps it is the sum of all parts I see that I find better. I also think that not using proper shooting technique very quickly bridges the gap between these cameras. A properly focused GXR shot does look brilliant to me.
Perhaps it is not accurate to describe it as clarity but IMO there is a positive side to omitting the AA-filter. (A negative side too, but that's another story.)
My point is, the difference is small, but it is there. And I don't see the artifacts you are talking about on same shots. When it comes to very small detail on distant landscape shots (grass/leaves) all APS-C cameras suffer from lack of resolution and cannot resolve that minute detail without introducing artifacts. However the GXR M does resolve minute detail that the 16MP sensor could not cope with, despite the MP advantage. It was a "Euro" sticker on a car where the GXR M would resolve - barely - the stars on the sticker as points but where the Sony would not even resolve the stars at all, just a smear, that could not be recovered after sharpening. And that is not false detail, that is real detail lost.
Technically they are so close I'd call it a wash, except for the corner/smear issues. But there are instances where I have seen the benefits of the AA-filter less sensor. As, I said, I don't care whether a sensor has AA-filter or not. I think my Sony A900 was excellent, and couldn't care less about it having a filter or not. Also I think the 16MP Sony sensor has the best potential among the current APS-C sensors, it's just that they haven't managed to get the best out of it yet for our little niche.
The way I see it the more you sharpen an image that's hobbled by an AA filter the more strained the image look. The Ricoh is already nearly there without the need for sharpening.
Douglas, I would love to know more about the AA-filter vs. no AA filter discussion. All I can say is that the GXR is the first no-AA filter camera I have had (and I have had most of them) and I don't use any sharpening at all with these files, and that's a first. I have processed tens of thousands of images from Nikon's and Sony's best, and they all need some sharpening to look their best. For my work, the combination of the CV UWAs on that sensor works.
If you can point me to theSuede's discussions on this, I'd be grateful or, even better, please ask him to join this thread. I have sold all my other gear and am using this ugly little thing because it makes the best files I have seen, especially for my interiors.
Of particular interest are these other artifacts you mention; more on that please. Lloyd pointed out the colour shift in the corners in my files (subtle, but there) and then I realises that there are corrections available in-camera that I will be playing with. But, AFAIK, this colour shift is more a property of using RF lenses; the Leica M8/9 has the same problems I believe (I have never used one).
Anyway, always ready to learn, so let's hear more! Cheers, KL
bluetsunami wrote:
The way I see it the more you sharpen an image that's hobbled by an AA filter the more strained the image look. The Ricoh is already nearly there without the need for sharpening.
The point is that having no AA filter hobbles a digital sensor in a different way, by actually adding digital artifacts that can be misinterpreted as more detail. It's also partially why AA-less cameras initially seem noisier. If adding sharpening looks more strained, than it's just a good job of sharpening. There's no free lunch, and it's why most DSLR makers don't yank the AA filter (which would be cheaper,) because they're less concerned with edge performance, although I might argue that DSLRs would also be better off at the edges without an AA filter.
Hi, Kit. You could just PM Joakim "theSuede" and ask, because his threads are scattered all over. He actually knows what he's talking about, compare to me.
kosmoskatten wrote:
On LCD cameras I find I struggle to cope with those speeds and I find that with anything above 50mm on the NEX5n and the GXR I have a much worse keeper ratio than before.
Funny you should write this; I was sorting through a bunch of photos I took with the CV75 last weekend - indoor stadium types - and am not very happy with my keeper ratio. Before coming here I was musing about making the 50mm my longest lens on this camera, at least for everyday "sans tripod" use.
You know plenty, Douglas! I will PM him. From a shooter's perspective, though, if no AA-fliter does add artifacts and the end result is that these artifacts look good the the experienced eye, then not a problem. Cheers, Kl
Can't argue with that, Kit! Just trying to dispell the idea that no AA filter is somehow more representative of the scene, and that AA-filtered can't get there with the right processing. I'd still prefer no AA to sharpen up my corners.
Douglas; I am of course interested in the technical side of things but I am with Kit and I think that if there is no smoke I won't look for a fire.
I think I have a clue as to what kind of artifacts that _might_ be referred to, but from my experience those artifacts are simply not visible in print, unless you exacerbate them with poor post pro. Might see them at 100% view on screen if you know what you are looking for, but from a printer's perspective I can totally dismiss them. The slight ink diffusion on paper that takes place when printing will mask this.
That is also why I find that the difference between NEX5n output, GXR M output and a whole lot of other cameras output is negligible when it comes to the final print. The exception being smear, excessive moir� and color casts that are not addressed in post pro.
What I don't find negligible is the amount of time and effort for me to get the file ready for printing - the way I like it. Here lies the forte of the GXR M mount, as well as with the NEX5n, I find both cameras easy to post process into print ready files of good quality.
I just slightly prefer the GXR M files.
PS:
Douglas, we might be talking in circles here.
In my opinion a well designed AA filter on a great sensor coupled to an awesome lens will yield a perfectly fine print for the same reasons and will in most cases be indistinguishable from an AA-filter less camera print.
Howdy. I've been typing on my little smart phone, so maybe I haven't been clear.
I'm not trying to imply that an AA-less camera is bad. I'm just saying the artifacts that cause more perceived resolution leads shooters to believe that there is more pop and detail than with an AA-filtered camera, and the camera has some magic (M9 shooters?? ,) but you can add sharpening and artifacts to an AA-filtered camera to get similar results. I often see comparisons between AA and AA-less cameras where the sharpening is the same, which is like putting the same fuel in two wildly different cars (terrible analogy.)
Since I'm using rangefinder lenses, I'd be happy with the trade offs of using no AA filter, and I'm still interested to see what happens if one is yanked of the NEX-7 or 5N. It'll be interesting to see what trade offs the Fuji X-Pro1 is making, with it's new CFA.
In the spirit of this thread title, despite currently being a NEX owner, I should make it clear that I have love for the Ricoh GXR M module.
douglasf13 wrote:
The point is that having no AA filter hobbles a digital sensor in a different way, by actually adding digital artifacts that can be misinterpreted as more detail. It's also partially why AA-less cameras initially seem noisier. If adding sharpening looks more strained, than it's just a good job of sharpening. There's no free lunch, and it's why most DSLR makers don't yank the AA filter (which would be cheaper,) because they're less concerned with edge performance, although I might argue that DSLRs would also be better off at the edges without an AA filter.
The thing is, how does one even differentiate between whats false detail and actual detail? And what bearing does it have on real world images and the very real difference in sharpness at the pixel level? No amount of sharpening will recreate the detail (be it false or real) the GXR is picking up. All I know is that the detail from RAW files from the Ricoh is phenomenal and takes really well to resizing and sharpening while retaining detail. One area being fine texture of objects which gives various lenses an almost Zeiss like feel (versus files from AA'ed cameras) due to this.
And I'd rather have a bit of noise that can be cleaned up in post to the specified strength I need than a catch-all reduction in detail to protect against "false detail" and moire. This is a proof is in the pudding situation for me after playing with many GXR DNGs. I just prefer the quality of files from the GXR vs what I've played with from the 5N.
I'm personally surprised no one as of yet has gotten the AA filter taken out of the 5N (or better yet the NEX7). Interested in seeing what effects it has on various problem lenses.
bluetsunami wrote:
The thing is, how does one even differentiate between whats false detail and actual detail? All I know is that the detail from RAW files from the Ricoh is phenomenal and takes really well to resizing and sharpening while retaining detail. One area being fine texture of objects which gives various lenses an almost Zeiss like feel due to this.
And I'd rather have a bit of noise that can be cleaned up in post than a reduction in detail to protect against "false detail" and moire. This is a proof is in the pudding situation for me after playing with many GXR DNGs. I just prefer the quality files from the GXR vs what I've played with from the 5N....Show more →
Hi, bluetsunami. That's my point. Many love the way no AA filter looks. I'm just saying it's possible to get that look with an AA filter, too. I'd certainly be fine with the no AA filter look, and the improved corners would be great, too. Granted, shooting a model wearing finely patterned clothes used to drive me batty in the studio with my Leaf back, but I don't do that kind of work, anymore, and it had pretty large pixels.
To quote Joakim,
"And btw, RESOLUTION can never go up by getting rid of the AA filters. In fact it often goes DOWN in the purely factual sense when looking at a statistical "average" scene - due to a larger number of situations where the interpolation fails to get the missing two colours per pixel right.
What you DO get is a higher contrast at the frequencies close to Nyquist, which can be mistaken for "higher resolution", especially since the main "benefit" you get from it is a very much stronger luma+chroma noise at the pixel level (100% Nyquist) in the interpolation stage of the raw conversion. It's quite easy to add a light sprinkle of "perfect" noise to a normal image - most people will see that as "increased detail". "
Joakim probably knows the deal with the GXR microlenses. We should ask him. 
). 
,) but you can add sharpening and artifacts to an AA-filtered camera to get similar results. I often see comparisons between AA and AA-less cameras where the sharpening is the same, which is like putting the same fuel in two wildly different cars (terrible analogy.)